To be a good citizen on the image orchestration protocol, your image must respond to health check queries. When an image fails to respond to health check with an appropriate exit code, it will be restarted. Images that fail to eventually start and reach a healthy state are at risk of non execution.
To support monitoring by the community, your service should maintain a local status.json file that shares its current status with the outside world. The contents of this file are publicly available from every node running the service via the network status page.
The following Dockerfile example creates a Node.js service based on Alpine Linux. The service uses a shell script that executes a Node.js script for performing the health check and updating status.json periodically.